采用门控循环单元神经网络和多特征融合的铣削刀具磨损监测  

Milling Tool Wear Monitoring by Using Gated Recurrent Unit Neural Network and Multi-feature Fusion

作　　者：葛慧 韩林池 麻俊方 宋清华[2] 王润琼 刘战强[2] 杜宜聪 王兵[2] 蔡玉奎[2] 赵金富 GE Hui;HAN Linchi;MA Junfang;SONG Qinghua;WANG Runqiong;LIU Zhanqiang;DU Yicong;WANG Bing;CAI Yukui;ZHAO Jinfu(Sinotruk Jinan Power Co.,Ltd.,Ji'nan 250220,China;School of Mechanical Engineering,Shandong University,Ji'nan 250061,China)

机构地区：[1]中国重汽集团济南动力有限公司,济南250220 [2]山东大学机械工程学院,济南250061

出　　处：《机械科学与技术》2024年第4期667-673,共7页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金项目(51922066,51875320)。

摘　　要：为实现汽车发动机缸盖生产中刀具磨损状态的监测,提高刀具磨损监测方法的计算效率和识别精度,基于门控循环单元神经网络和多特征融合方法提出了面向铣刀后刀面磨损带宽度识别的刀具状态监测方法。通过铣削力信号数据对所提出方法的有效性进行了验证,分析了不同超参数设置对模型识别精度的影响机制,给出了最优超参数,实现了对铣削刀具磨损的精确识别。To realize the tool wear condition monitoring in the production of a vehicle engine's cylinder head and to enhance the computational efficiency and recognition accuracy of tool wear monitoring,a tool condition monitoring method based on the gated recurrent unit neural network and the multi-feature fusion method is proposed for identifying the width of milling tool flank wear.The effectiveness of the proposed method is verified with the milling force signal data,and the effects of different hyper-parameter settings on the model recognition accuracy is analyzed.The optimal hyper-parameters are given;the accurate recognition of milling tool wear is realized.

关 键 词：刀具磨损 铣削力信号 状态监测 门控循环单元神经网络 

分 类 号：TH164[机械工程—机械制造及自动化]